Limited linguistic inclusivity in public health communication leaves many South African communities underserved, particularly regarding critical information on zoonotic diseases such as rabies. This pilot study addresses this gap by developing and evaluating AI-driven methods for delivering reliable rabies information to Sepedi speakers, a low-resource language group. The study presents a novel, curated Sepedi dataset of 60 question–answer pairs, created through a systematic pipeline: thematic analysis of authoritative English sources guided the synthetic generation of QA pairs, which were then translated and manually verified by a native-speaking expert. This dataset was used to compare two large language models, GPT-4o and Gemini-1.5 Flash, under both base and fine-tuned conditions. Evaluation used a human-centred rubric assessing fluency, accuracy, and cultural appropriateness. The findings reveal a key nuance in applying LLMs to low-resource domains. The base GPT-4o model, with strong foundational multilingual capabilities, outperformed all other configurations, including its own fine-tuned variant.
In contrast, fine-tuning provided a marked improvement for the less capable base Gemini model. This result indicates that fine-tuning can enhance weaker models; its benefits are not universal and may be outweighed by the strong zero-shot performance of state-of-the-art architectures when training data is scarce. The curated Sepedi rabies QA dataset will be released under an open licence to support future work in low-resource public health communication.

The critical lack of structured terminological data for South Africa’s official languages hampers progress in multilingual NLP, despite the existence of numerous government and academic terminology lists. These valuable assets remain fragmented and locked in non-machine-readable formats, rendering them unusable for computational research and development. Mafoko addresses this challenge by systematically aggregating, cleaning, and standardising these scattered resources into open, interoperable datasets. We introduce the foundational Mafoko dataset, released under the equitable, Africa-centered NOODL framework. To demonstrate its immediate utility, we integrate the terminology into a Retrieval-Augmented Generation (RAG) pipeline. Experiments show substantial improvements in the accuracy and domain-specific consistency of English-to-Tshivenda machine translation for large language models. Mafoko provides a scalable foundation for developing robust and equitable NLP technologies, ensuring South Africa’s rich linguistic diversity is represented in the digital age.

This paper aims to provide a brief exploration of two versions of Creative AI, namely the prompting of portraits by using AI text-to-image generators and the use of GAN, AICAN and Facer to create AI generated portraits. These two versions are in turn compared to corresponding debates in the field of art history, namely the image-text debate as positioned by the image scholar, WJT Mitchell, followed by the concept of schemata as proposed by the art historian EH Gombrich. First, Mitchell’s understanding of the nature of the image versus text is utilized to compare portraits prompted through text-to-image generators. Secondly, Gombrich’s schemata is compared with recent AI portraits generated by means of image datasets. The differences between the art historical and the Creative AI processes are explored to draw initial conclusions about the future of portraiture and creativity.

The Covid-19 pandemic has resulted in the worst downturn in the global economy since the Great Depression in the 1930s. To face the challenges of the global economy, a person needs to possess basic skills including educational skills. Education plays a vital role in building a competitive economy that will hardly be affected by crisis and will be able to ensure that there are high rates of social development. The student population has become very diverse over the decades, making it difficult to teach. Teaching has become very complex to handle because of the increase in a variety of teaching strategies and the diverse student population. There is therefore a need for inclusive and equity pedagogy where teaching considers the diversity of students and the need for teachers to develop teaching strategies that support all students, especially those from disadvantaged backgrounds. The most expensive education is the one that is not completed. This conceptual paper looks at the importance of the Learner Management System (LMS) in implementing data-driven instruction to achieve quality education for all types of students. The LMS is a software system that tracks students’ participation and progress through data systems and assessments. It’s a platform that stimulates an environment for learner achievement and engagement. ‘Data-driven instruction’ can be defined as using student data to enhance instructional practices in the classroom to address the needs and learning styles of individual students. Additionally, data-driven instruction will be explored to discover how it can be used as a systematic and purposeful work to maximise the students’ performance. The study will provide recommendations on how LMS and data-driven instruction can be used to give direction to decisions to improve the students' outcomes.